From 62462b48bc8d55aec8c6d321c07f9d1d9d6885b4 Mon Sep 17 00:00:00 2001
From: Strahinja Val Markovic <val@markovic.io>
Date: Fri, 30 Aug 2013 14:38:23 -0700
Subject: [PATCH] Ident collector handles escaped starting quotes

Previously, it would consider \"foo\"bar" as a slash and a "foo\"bar" string and
this would screw up tracking of quotes in the file.

Fix #535.
---
 cpp/ycm/IdentifierUtils.cpp            | 17 +++++++++++------
 cpp/ycm/tests/IdentifierUtils_test.cpp | 14 ++++++++++++++
 2 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/cpp/ycm/IdentifierUtils.cpp b/cpp/ycm/IdentifierUtils.cpp
index e2d41deb..e35ed07a 100644
--- a/cpp/ycm/IdentifierUtils.cpp
+++ b/cpp/ycm/IdentifierUtils.cpp
@@ -37,13 +37,18 @@ const char *COMMENT_AND_STRING_REGEX =
   "|"
   "/\\*.*?\\*/"  // C-style comments, '/* ... */'
   "|"
-  // Anything inside single quotes, '...', but mind the escaped quote and the
-  // escaped slash (\\)
-  "'(?:\\\\\\\\|\\\\'|.)*?'"
+  // Anything inside single quotes, '...', but mind:
+  //  1. that the starting single quote is not escaped
+  //  2. the escaped slash (\\)
+  //  3. the escaped single quote inside the string
+  // "(?<!\\\\)'(?:\\\\\\\\|\\\\'|.)*?'"
+  "(?<!\\\\)'(?:\\\\\\\\|\\\\'|.)*?'"
   "|"
-  // Anything inside double quotes, "...", but mind the escaped double quote and
-  // the escaped slash (\\)
-  "\"(?:\\\\\\\\|\\\\\"|.)*?\"";
+  // Anything inside double quotes, "...", but mind:
+  //  1. that the starting double quote is not escaped
+  //  2. the escaped slash (\\)
+  //  3. the escaped double quote inside the string
+  "(?<!\\\\)\"(?:\\\\\\\\|\\\\\"|.)*?\"";
 
 const char *IDENTIFIER_REGEX = "[_a-zA-Z]\\w*";
 
diff --git a/cpp/ycm/tests/IdentifierUtils_test.cpp b/cpp/ycm/tests/IdentifierUtils_test.cpp
index f673f8be..7dd7d8c8 100644
--- a/cpp/ycm/tests/IdentifierUtils_test.cpp
+++ b/cpp/ycm/tests/IdentifierUtils_test.cpp
@@ -112,6 +112,20 @@ TEST( IdentifierUtilsTest, RemoveIdentifierFreeTextWorks ) {
                 "foo \n"
                 "bar \n"
                 "qux " );
+
+  EXPECT_STREQ( RemoveIdentifierFreeText(
+                  "\\\"foo\\\""
+                  "'\"'"
+                  "'bar' zoo'test'"
+                ).c_str(),
+                "\\\"foo\\\" zoo" );
+
+  EXPECT_STREQ( RemoveIdentifierFreeText(
+                  "\\'foo\\'"
+                  "\"'\""
+                  "\"bar\" zoo\"test\""
+                ).c_str(),
+                "\\'foo\\' zoo" );
 }